Evaluation of Visual Comfort for Wallpaper by Image Analysis

The purpose of this study is to evaluate visual comfort of wallpaper by using image analysis. First of all, 30 human observers estimated the visual comfort for various samples of wallpaper on a CRT display. Next, we examined the relationship between the visual comfort estimated by the human observers and each of 4 feature parameters obtained by image analysis, that is, average lightness, average chromaticness, average spatial frequency, and entropy in spatial frequency domain. Clearly, the visual comfort of wallpaper has a good positive correlation with the average lightness, and a good negative correlation with the average chromaticness. Furthermore, a multiple regression analysis was performed by using these feature parameters to predict the visual comfort. The regression equation included 6 explanatory variables, that is, the 4 feature parameters and the second order terms of average spatial frequency and entropy. The result showed that the regression coefficients were negative for the second order terms of the 2 feature parameters, therefore, wallpaper with medium values of average spatial frequency and entropy was evaluated as comfortable. In other words, this suggests that the wallpaper with medium coarseness (fineness) and moderate spatial periodicity tends to bring the visual comfort.